.inp file for groups and individual covariates

[giulia.sozio]

Hello,
I’ve a question on the codification of groups in the .inp file. I've searched a lot both in the manual and in the forum but I could not find the solution. I’ve found that there are 2 ways to code groups and individual covariates:
1. Using 2 columns representing the frequencies depending on group: “1 0” indicating group 1, and “0 1” indicating group 2, followed by the value of the covariate:
11111111 1 0 123.211;
11111111 0 1 92.856;
11111110 1 0 122.115;
11111110 1 0 136.460;
2. Using a column of 1’s, indicating the frequency for each individual, followed by a column containing a 0/1 dummy code to indicate group:
11111111 1 1 123.211;
11111111 1 0 92.856;
11111110 1 1 122.115;
11111110 1 1 136.460;

In several parts of the manual I found .inp files formatted with a column of 1’s (hence as in the second way). But in the forum is often suggested to format the inp.file as in the first way.
My question is: what is the difference between these two methods? I’ve tried to use them both, and I obtained different results! Which one should I use?
I copy below my results (I’ve used CJS just to try, then I will use the robust design): note that in the real function parameters some values are identical or similar between the two methods (but those of different groups). Can anyone help me?

Thank you very much,
Giulia


First:

SIN Link Function Parameters of {p(,)phi(.)}
95% Confidence Interval
Parameter Beta Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 58.367178 0.0299124 58.308550 58.425807
2:Phi -48.487191 0.0300535 -48.546096 -48.428287
3:p 6.0671631 0.1410902 5.7906262 6.3437000
4:p 9.5299983 0.1392422 9.2570836 9.8029130


Real Function Parameters of {p(,)phi(.)}
95% Confidence Interval
Parameter Estimate Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 0.9847377 0.0036671 0.9756069 0.9904840
2:Phi 0.9892750 0.0030956 0.9811559 0.9939177
3:p 0.3928270 0.0689055 0.2686115 0.5326510
4:p 0.4474869 0.0692360 0.3187125 0.5837131


Second:

SIN Link Function Parameters of {p(.)phi(.)}
95% Confidence Interval
Parameter Beta Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 1.8005213 0.0211315 1.7591035 1.8419392
2:Phi 1.8185108 0.0299119 1.7598835 1.8771382
3:p 31.253913 0.0993921 31.059104 31.448721
4:p 3.3576145 0.1410878 3.0810824 3.6341466


Real Function Parameters of {p(.)phi(.)}
95% Confidence Interval
Parameter Estimate Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 0.9868645 0.0024059 0.9812096 0.9908335
2:Phi 0.9847377 0.0036670 0.9756070 0.9904839
3:p 0.4193470 0.0490453 0.3273397 0.5173261
4:p 0.3928272 0.0689043 0.2686136 0.5326487

[cooch]

Hello,
I’ve a question on the codification of groups in the .inp file. I've searched a lot both in the manual and in the forum but I could not find the solution. I’ve found that there are 2 ways to code groups and individual covariates:
1. Using 2 columns representing the frequencies depending on group: “1 0” indicating group 1, and “0 1” indicating group 2, followed by the value of the covariate:
11111111 1 0 123.211;
11111111 0 1 92.856;
11111110 1 0 122.115;
11111110 1 0 136.460;
2. Using a column of 1’s, indicating the frequency for each individual, followed by a column containing a 0/1 dummy code to indicate group:
11111111 1 1 123.211;
11111111 1 0 92.856;
11111110 1 1 122.115;
11111110 1 1 136.460;

In several parts of the manual I found .inp files formatted with a column of 1’s (hence as in the second way). But in the forum is often suggested to format the inp.file as in the first way.
My question is: what is the difference between these two methods? I’ve tried to use them both, and I obtained different results! Which one should I use?
I copy below my results (I’ve used CJS just to try, then I will use the robust design): note that in the real function parameters some values are identical or similar between the two methods (but those of different groups). Can anyone help me?

Thank you very much,
Giulia


First:

SIN Link Function Parameters of {p(,)phi(.)}
95% Confidence Interval
Parameter Beta Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 58.367178 0.0299124 58.308550 58.425807
2:Phi -48.487191 0.0300535 -48.546096 -48.428287
3:p 6.0671631 0.1410902 5.7906262 6.3437000
4:p 9.5299983 0.1392422 9.2570836 9.8029130


Real Function Parameters of {p(,)phi(.)}
95% Confidence Interval
Parameter Estimate Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 0.9847377 0.0036671 0.9756069 0.9904840
2:Phi 0.9892750 0.0030956 0.9811559 0.9939177
3:p 0.3928270 0.0689055 0.2686115 0.5326510
4:p 0.4474869 0.0692360 0.3187125 0.5837131


Second:

SIN Link Function Parameters of {p(.)phi(.)}
95% Confidence Interval
Parameter Beta Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 1.8005213 0.0211315 1.7591035 1.8419392
2:Phi 1.8185108 0.0299119 1.7598835 1.8771382
3:p 31.253913 0.0993921 31.059104 31.448721
4:p 3.3576145 0.1410878 3.0810824 3.6341466


Real Function Parameters of {p(.)phi(.)}
95% Confidence Interval
Parameter Estimate Standard Error Lower Upper
------------------------- -------------- -------------- -------------- --------------
1:Phi 0.9868645 0.0024059 0.9812096 0.9908335
2:Phi 0.9847377 0.0036670 0.9756070 0.9904839
3:p 0.4193470 0.0490453 0.3273397 0.5173261
4:p 0.3928272 0.0689043 0.2686136 0.5326487

Chapter 11 - section 11.7

Using individual covariates to code for 'groups' will change only the beta terms (relative to a model where groups aare coded using column frequencies). If the real estimates differ, then either (i) your individual covariate-based analysis is set up incorrectly, or (ii) the frequency column approach is coded incorrectly, or (iii) both are coded incorrectly. If you set things up correctly, you should reconstitute the same real parameter estimates, regardless of your choice of 'input file coding schemes'.